Josephine Akosa

Josephine Akosa is a Assistant Research Professor of IT, Analytics, and Operations in the Mendoza College of Business at the University of Notre Dame. Her research interest is in high-dimensional statistical inference. Rapid advancement in technology has allowed an accelerated increase in the amount of data collected in many research fields, thereby allowing for thousands of hypotheses to be tested with a relatively small number of experimental units. The problem here is that multiplicity adjustment must be made but traditional methods are designed in the realm of independent and normally distributed data with the dimension of the data being smaller than the sample size, when in fact these are not the characteristics of high-dimensional inferential problems. Professor Akosa focuses on developing powerful multiple comparison procedures for non-normally distributed data that incorporate small sample adaptations to estimation procedures of model parameters. She has also worked on projects involving imbalanced data modeling.

Research Interests

Multiple Testing; High Dimensional Data; Imbalanced Data Analysis; Data Mining

Peer Reviewed Conference Proceedings

• Akosa, J. S. Predictive Accuracy: A Misleading Performance Measure for Highly Imbalanced Data. SAS Institute Inc. 2017. Proceedings of the SAS® Global Forum 2017 Conference. Cary, NC: SAS Institute Inc.
• Akosa, J. S., Kelly. S. Applications of Data Mining Techniques in Improving Breast Cancer Diagnosis. SAS Institute Inc. 2016. Proceedings of the SAS® Global Forum 2016 Conference. Cary, NC: SAS Institute Inc.

Non-Peer Reviewed Publications

• Akosa, J. S. " Resampling-based multiple comparisons for generalized linear models", Master's thesis, The University of Texas at El Paso.

Poster & Conference Presentations

• J. S. Akosa, M. McCann. “Learning from Imbalanced data: A review of some existing methodologies”. 2017 Joint Statistical Meeting, July 29 – Aug. 3, 2017: Baltimore, Maryland.
• J. S. Akosa. “Predictive Accuracy: A Misleading Performance Measure for Highly Imbalanced Data”. SAS Global Forum 2017, April 2 – 5, 2017: Orlando, Florida.
• J. S. Akosa, S. Kelly. “Applications of Data Mining Techniques in Improving Breast Cancer Diagnosis”. SAS Global Forum 2016, April 18 – 21, 2016: Las Vegas, Nevada.
• J. S. Akosa, S. Kelly. “Applications of Data Mining Techniques in Improving Breast Cancer Diagnosis”. 2016 SAS Analytics Day at the Oklahoma State University, April 26, 2016: Stillwater, Oklahoma.
• J. S. Akosa, S. Kelly. “Improving Breast Cancer Diagnosis via Data Mining Techniques”. Analytics 2015, Oct. 26 – 27, 2015: Las Vegas, Nevada.
• J. S. Akosa, D. Contreras, M. Diaw, K. Rudra. “‘PQR’ Entertainment Marketing Analysis”. 2015 SAS Analytics Day at the Oklahoma State University, May 3, 2015: Stillwater, Oklahoma.
• J. S. Akosa, A. Wagler. “Resampling-based multiple comparisons for generalized linear models”. Southern Regional Council on Statistics 2014 Summer Research Conference, June 1 – 4, 2014: Galveston, Texas.

Awards

• "Vik Family Outstanding Graduate Student Award", Department of Statistics, Oklahoma State University, 2018
• "Statistics Student Travel Award", Department of Statistics, Oklahoma State University, 2018
• "Statistics Student Travel Award", Department of Statistics, Oklahoma State University, 2017
• "Finalist, OSU Health Data Shootout", Center for Health Systems Innovation, Oklahoma State University, 2016
• "Graduate and Professional Student Association Travel Award", Graduate School, Oklahoma State University, 2016
• "Robert Morrison Scholarship", Department of Statistics, Oklahoma State University, 2015
• "Student Travel Award", College of Science, University of Texas at El Paso, 2014
• "Student Travel Award", Mathematical Science Department, University of Texas at El Paso, 2014
• "Summer Research Conference Student Travel Award", Southern Regional Council on Statistics, 2014